Experiential v0.7.149 发布：付费响应恢复与中断 rollout 处理更新

Official release notes (https://github.com/experientiallabs/experiential/releases/tag/v0.7.149) Completed responses with missing price-relevant usage and interrupted paid cells can recover as fresh rollout generations in explicitly uncapped evaluations. Saved responses, request identities, receipts, and unknown liabilities remain intact. Recovery does not resend a saved HTTP request or invent zero usage. The SDK saves paid responses before reporting invalid usage or a charge beyond the reserved request bound. Those validation errors remain nonretryable. Raw bodies with truncated tool JSON are also saved before the parsing error reaches the caller, and exact replay makes no new HTTP request. Length-limited responses with incomplete tool-call JSON are classified as interrupted responses and can use the same bounded rollout recovery. Ordinary malformed responses retain their existing handling. The existing three-generation limit remains separate from the requested repeat count. Finite budgets do not authorize recovery with unknown prior liability. Completed evidence remains available for free replay. Reports retain unknown aggregate experiment spend where necessary and calculate model comparison costs from valid rollouts. Includes PR #1185 and the previously released reasoning display changes. The native gateway dependency is `exp-gateway-native==0.3.129`. Repository README (current repository documentation, not release changes) # Experiential [](https://github.com/experientiallabs/experiential/actions/workflows/gateway-latency.yml?query=branch%3Amain) Experiential is an open source gateway and router for agent workflows: 1. Use hosted, BYOK, and local models through one OpenAI-compatible API. 2. Control which users and agents can use which models, for which use cases, and how much they can spend. 3. Turn production traffic into a custom router or model optimized for quality, speed, and cost.  🌐 Platform | 📚 Docs | Discord ## Getting Started Start a local OpenAI-compatible gateway. On first run, the setup wizard uses the shared provider, model, and reasoning-effort selectors, persists every selected provider connection, then shows defaults for the public alias, identity, and `$50.00` command budget before printing a one-time key: ```bash pip install experiential exp ``` Choose a public alias such as `opus-5`, capture the issued key, and send a request: ```bash export EXP_GATEWAY_KEY=... curl http://127.0.0.1:8000/v1/chat/completions \ -H "Authorization: Bearer $EXP_GATEWAY_KEY" \ -H 'Content-Type: application/json' \ -d '{"model":"opus-5","messages":[{"role":"user","content":"Help me"}]}' ``` ## Setup / get started with the hosted gateway Prefer a managed gateway to running one locally? The hosted platform at [platform.experientiallabs.ai](https://platform.experientiallabs.ai) serves the same OpenAI-compatible (and Anthropic Messages) API at `https://api.experientiallabs.ai/v1`. See [SETUP.md](./SETUP.md) for copy-paste prompts you hand to your coding agent (Claude Code, Cursor, Codex, and similar); the agent runs the setup for you. It collects four prompts: - Upload your LLM traces as telemetry: create an account instantly from your email, then pull or upload your existing LLM traces onto the platform as telemetry. - Connect your inference provider keys (BYOK): create an account, then connect your own OpenAI, Anthropic, Gemini, Azure, Bedrock, Fireworks, or OpenRouter keys for free pass-through. - Start calling models on the gateway: make your first `/v1` call with the OpenAI and Anthropic SDKs using an `xpl_` key, and optionally repoint your existing coding agents. - Full onboarding: create an account instantly from your email, connect your keys, import your spend, then repoint every coding agent (Claude Code, Cursor, Codex, Aider, and similar) or Conductor at the gateway. ## Using the API Start the local gateway with `exp` (or `exp run`); the compiled native data plane serves every route on loopback. From Python, load a fitted project router as an official OpenAI client backed by its own private gateway: ```python import exp with exp.load_router("my-project") as client: response = client.chat.completions.create( model="my-project", messages=[{"role": "user", "content": "hello"}], ) ``` ## Capture Traffic Capture is designed to collect supported traces from the Codex and Claude Code desktop apps using your existing ChatGPT or Claude subscription, while requests go directly to OpenAI or Anthropic. It requires macOS, Python 3.13 or newer, and approval for its extension and certificate: ```bash exp capture ``` Capture is experimental; macOS network reliability is still under investigation. Use Experiential's gateway as middleware to control costs, enforce guardrails, and optimize quality, costs, and speed. Leave the terminal open while you work. Press Ctrl+C to stop capture. View your [capture telemetry](https://platform.experientiallabs.ai/api-keys?section=capture). ## Optimize from Traffic First, collect OpenTelemetry traces from your current agent. If you just want to try it out, grab the public [terminal-tasks OTLP dataset](https://huggingface.co/datasets/experiential-labs/wmo-terminal-tasks-traces): ```bash curl -L -o traces.otel.jsonl \ https://huggingface.co/datasets/experiential-labs/wmo-terminal-tasks-traces/resolve/540883e451dc13d34fb50fdd36b143cb0f1fb0db/traces.otel.jsonl ``` Then build a project. The build command walks you through providers, models, and budget, and asks for your trace file: ```bash # Build simulation from your agent traces and optimize a router against it exp build support-agent ``` After collecting traces from your router, fine-tune an open source model you own using [Tinker](https://tinker.thinkingmachines.ai/). ```bash exp optimize model support-agent ``` ## Telemetry Anonymous aggregate PostHog product telemetry is enabled by default. It never includes prompts, traces, actions, observations, paths, model names, credentials, or raw customer content. ```bash exp config telemetry status exp config telemetry disable exp config telemetry enable ``` The preference is stored locally in `.exp/settings.toml`. ## Development ```bash uv sync --extra dev uv run ruff format --check . uv run ruff check . uv run ty check uv run pytest -q ``` Repository and documentation conventions live in [AGENTS.md](./AGENTS.md).